Reporting in the journal Biological Psychiatry, researchers write of modeling schizophrenia in a computerized simulation of the brain's connections, called a "neural network." Yale psychiatrist Dr. Ralph Hoffman, an author on the paper, discusses what his team has learned from the model.
Copyright ? 2011 National Public Radio?. For personal, noncommercial use only. See Terms of Use. For other uses, prior permission required.
IRA FLATOW, host:
Great mathematician John Nash, a subject of a movie, "A Beautiful Mind," once described his delusions from schizophrenia in this way: quote, "I started to see crypto-communists everywhere. I started to think I was a man of great religious importance and to hear voices all the time. I began to hear something like a telephone call in my head from people opposed to my ideas. The delirium was like a dream from which I seemed never to awake."
Scientists have observed these sorts of symptoms in many schizophrenic patients. But the big mystery is, what exactly causes the hallucinated conversations, where the tendencies schizophrenics have to slip from one topic to another in midsentence or to mix themselves up in these intricately imagined plots? How do you figure all that out?
Well, one way is to look at a bunch of computer patients, computers programmed to have a similar sort of behavior. And that's what my next guest has done. And his cyber-tinkering has allowed him to derive some new theories about schizophrenia. That research appears in the journal, Biological Psychiatry.
Dr. Ralph Hoffman is a professor of psychiatry at Yale University School of Medicine in New Haven, and he joins us by phone. Welcome to SCIENCE FRIDAY.
Dr. RALPH HOFFMAN (Professor of Psychiatry, Yale University School of Medicine): Oh, thank you very much. Very pleased to speak with you here.
FLATOW: What make you try to create a model on a computer about schizophrenia?
Dr. HOFFMAN: Well, primarily because these are such vexing manifestations of schizophrenia that are so - still so poorly understood. And then along with that, there are so many theories and ideas about what might be causing these problems. So what we ended up doing essentially is developing an artificial neural system implemented on a computer, an artificial brain simplified but basically - and then inducing a wide range of hypothesized illness mechanisms and to see which one actually seemed to produce the better fit in terms of what we are seeing clinically and experimentally with actual patients.
FLATOW: And so do you have any idea what's behind the, you know...
(Soundbite of laughter)
Dr. HOFFMAN: Well, that's the big question.
(Soundbite of laughter)
Dr. HOFFMAN: We simulated about eight different core disturbances and -with this system and compared it, as I said, to human behavior. And both the artificial neural systems and the humans were given tasks of remembering and recalling stories. So - and actually, what surfaced was a clear cut, at least a statistical winner, in terms of the problem that did the best in terms of simulating these patients. And that was what we now term as hyperlearning.
FLATOW: So you had some - you had one.
Dr. HOFFMAN: Yeah. We landed one.
(Soundbite of laughter)
FLATOW: Landed - hyperlearning, did you call it?
Dr. HOFFMAN: Yes, hyperlearning.
FLATOW: And what does that mean?
Dr. HOFFMAN: Well, what it means really is that in the artificial neural system that particular modules, networks within the system, were engaged in a learning process. So they were essentially fed stories of certain sorts and they had to learn the stories and put them in an abstract form that enabled the stories to be actually downloaded into essentially a permanent memory in this artificial system. And then the system, it kind of unwound that process in the other direction, was able to take these memory traces and recreate the stories.
So what we found is essentially is if this memory consolidation process was accelerated to an excessive degree that the system began to confuse different stories or memories and to corrupt these representations in very specific ways. It seemed to be suggestive of certain language manifestations of schizophrenia.
And the reason why this is of potential importance is that - in the real human brain, that when we have experiences interacting with others, seeing things on television, whatever, that the common way that we parse these experience and organize them and remember them for ourselves and to tell other people about it is through stories.
So stories are critical kind of organization for information, particularly social information, and really makes us who we are as individuals, the stories that we tell in some sense.
So these stories, like other episodic memories, when we experience them and register them in memory, take a long, long time to become incorporated into a permanent store. So - and we don't - so that - so what the study really suggests essentially - so we're talking weeks to months. So initially, when we have these experiences and have memories in our minds about what we've gone through and what we've experienced or seen that it's kind of a snapshot or a set of snapshots of these experiences. But then, gradually, these are incorporated into more abstract story structures that are related to other stories that we have learned and remembered. So there's a generalization process and an efficiency-making process.
FLATOW: Is there anyway that you can incorporate what you've learned from your computers into helping people?
Dr. HOFFMAN: Yeah. Well, let me - just to finish out this story. So what it really says is the reason why the human brain has to do this so slowly is that if it actually speeds it up that memories become intermingled and corrupted in certain ways. And it might actually -we're hypothesizing - produce the manifestations of what we call schizophrenia.
FLATOW: And as far as helping people now, you're using what you know?
Dr. HOFFMAN: Well, what we need to do is press on and we need to some, I mean, we have, I think, an interesting hypothesis and some preliminary supporting data, because this study was actually done in comparison to a group of actual patients.
And what we need to do now is to look directly in the brains of these individuals and to see if this hyperlearning process can be confirmed, not just through their manifest language behavior, but using certain kinds of neuro-imaging. So that's what we're pursuing now.
FLATOW: Well, thank you very much for taking time to be with us today. Good luck to you.
Dr. HOFFMAN: Oh, thank you.
FLATOW: Dr. Ralph Hoffman, professor of psychiatry at Yale University School of Medicine in New Haven. I'm Ira Flatow. This is SCIENCE FRIDAY from NPR.
Copyright ? 2011 National Public Radio?. All rights reserved. No quotes from the materials contained herein may be used in any media without attribution to National Public Radio. This transcript is provided for personal, noncommercial use only, pursuant to our Terms of Use. Any other use requires NPR's prior permission. Visit our permissions page for further information.
NPR transcripts are created on a rush deadline by a contractor for NPR, and accuracy and availability may vary. This text may not be in its final form and may be updated or revised in the future. Please be aware that the authoritative record of NPR's programming is the audio.
Source: http://www.npr.org/2011/05/13/136280444/studying-schizophrenia-by-giving-it-to-computers?ft=1&f=1019
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.